It's pretty uncommon (never?) for a current gen high performance SOC to be dual sourced on two different processes. No one is benchmarking two basebands from two different sources against each other because there is nothing to be learned. This is the current leading edge fabs (excluding Intel) with the exact same processor and there is an opportunity to learn a ton about those processes because of it.
It also explains the geekbench scores being all over the place.
The Geekbench scores are "all over the place" probably due to differing thermal conditions, battery life, and battery life when the tests were run.
You know that the 360 CPU was dual sourced at launch? Microsoft paid the extra money to run the design at both IBM and Chartered Semi, in case one of the two had trouble with yields. Part of the reason why they beat the PS3 to market.
And yet the A8 wasn't like that which also probably had different thermal conditions, battery life, and background usage.
Sure it did. I have a bunch of Geekbench 3 scores of my old iPhone 6 and I've gotten pretty different scores.
Yes, but there was no way for someone to benchmark the two and IIRC, the process at Chartered Semi was IBM's 90nm.
How?It's pretty uncommon (never?) for a current gen high performance SOC to be dual sourced on two different processes. No one is benchmarking two basebands from two different sources against each other because there is nothing to be learned. This is the current leading edge fabs (excluding Intel) with the exact same processor and there is an opportunity to learn a ton about those processes because of it.
It also explains the geekbench scores being all over the place.
How?
Remember the A5, and then the A5 again? They pretty much benched the same.
Different processes are going to have very different temperature profiles and as a result throttle in different ways. It's probably a safe bet that the TSMC chips get hotter faster and throttle more aggressively.
LOL, proof?
Not sure why this is LOL worthy? The early results (like the charts I posted earlier) seem to show that the A9 is more inconsistent than the A8. I'm sure we'll see more details and comparisons soon.
I have a 6S here and it's fast, haven't had any throttling issues or anything else, so it's not like I'm complaining. It's just logical that a chip built on Samsung's 14nm process, which is supposed to be superior to TSMC's 16nm, would perform better but by how much we have absolutely no idea. Like the Chipworks folks said, it provides a very interesting opportunity to compare the two since it hasn't happened recently as far as I can tell.
Why is Samsung's 14nm supposed to be "superior" to TSMC 16nm?
I recall reading an article that said the density for Samsung was significantly better than the labeling would imply, which would theoretically allow for more chips per platter and therefor cost savings (all else being equal) but it's also a matter of determining performance characteristics of the different processes and as far as that goes I haven't read much or seen a good comparison.
Apple had two different A5s in the iPad 2 with the same battery, with different battery life. Most people didn't know enough to care, or care enough to know. Not that you'd know anyway because you couldn't easily tell.There is still a lot of confusion whether APL0898 is 6S only and APL1022 is 6S Plus only. Chipworks does not categorically say that they found both these chips in the iPhone 6S only. iFixit's teardown shows APL0898 on 6S and APL1022 on 6S Plus. Anyway it make better logical sense to have a different chip power a different device than 2 chips powering the same device as you can keep the performance and battery life consistent. With 2 chips powering the same 6S we will see different power efficiency and thus different battery life. Thats a night mare for PR. People would not like to have an inferior chip while buying a 6S. As far as I have heard Samsung has better density which is proven now and TSMC has better transistor performance (power efficiency) and yields. TSMC's transistor performance advantage needs to be verified through independent benchmarking and the A9 provides a perfect opportunity to do so. The key factor here seems to be supply volume. TSMC did not have the supply volume available to completely supply A9. Samsung were first to high volume production and I think the 6S will use only APL0898. APL0898 can be produced at Samsung and GF and thus it makes sense that the higher volume 6S uses it.
Apple had two different A5s in the iPad 2 with the same battery, with different battery life.
Could you provide proof of that. Who were the manufacturers of the different A5 chips ?
Apple multi-sources displays and frankly, these are much more obvious to consumers than an AP.
Multi sourcing a SoC manufactured on a bleeding edge FINFET process when it has the ability to show significantly different power efficiency characteristics is doubtful. Apple would have to explain why few consumers are getting better battery life (btw i mean above the margin of error) compared to the rest. 6S using APL0898 (Samsung) and 6S Plus using APL1022 (TSMC) keeps the product behaviour consistent. All 6S will perform alike as will 6S Plus. Differences will be within margin of error (<=2%)
Multi sourcing a SoC manufactured on a bleeding edge FINFET process when it has the ability to show significantly different power efficiency characteristics is doubtful. Apple would have to explain why few consumers are getting better battery life (btw i mean above the margin of error) compared to the rest. 6S using APL0898 (Samsung) and 6S Plus using APL1022 (TSMC) keeps the product behaviour consistent. All 6S will perform alike as will 6S Plus. Differences will be within margin of error (<=2%)
Apple wouldn't have to explain anything here because the differences are probably minor, but even if they were say 5% different, nobody would even notice that. It's not like reviewers buy multiple iPhones, do tear downs to determine the components, and then bench them based on component mix. Hell, most reviewers don't even formally test battery life.
Basically, unless one was overheating and causing crashes or whatever, it's a complete non-issue.
10% higher speed at fixed power normally translates to 15-20% lower power at fixed speed. Thats a significant difference and something which would be easily discernible in terms of power consumption and thus battery life. Apple would be in a spot as to explain why certain consumers got much better battery life.